We analyzed the length of the CAG repeats of the androgen receptor gene in Indian women with breast cancer, and compared the data with that of other populations across the world in an attempt to find a potential pattern of association. The study was undertaken on 1,408 individuals comprising 747 breast cancer patients and 661 control individuals recruited from three southern states of India: Andhra Pradesh, Tamil Nadu, and Karnataka. The comparison revealed no difference in mean length of the repeat between cases and controls in any of the three groups or in the analysis of pooled data. No significant difference between pre- and post-menopausal cases in any of the three groups or in the analysis of pooled data was observed. Most of the studies to date support either positive association (longer repeats--increased disease risk) or no association, and only 2 out of 20 studies reported negative association (inverse correlation between repeat length and disease risk). Comparison of these data with those from other populations revealed several interesting facts. Particularly notable is that repeat length shows association with breast cancer risk in a population-specific manner with most of the studies on American and Canadian women showing positive association, whereas those on Australian and Israeli women showing no association. Only one study had been conducted on other populations including Asians/South Asians; this restricted us from finding any patterns of association in these populations.